Impact by pancreatic stellate cells on epithelial-mesenchymal transition and pancreatic cancer cell invasion: Adding a third dimension in vitro

Experimental Cell Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Contents lists available at ScienceDirect

Experimental Cell Research journal homepage: www.elsevier.com/locate/yexcr

Research Article

Impact by pancreatic stellate cells on epithelial-mesenchymal transition and pancreatic cancer cell invasion: Adding a third dimension in vitro Emelie Karnevi a,b,n, Ann H. Rosendahl b, Katarzyna Said Hilmersson a, Moin A. Saleem c, Roland Andersson a a b c

Lund University and Skåne University Hospital, Department of Clinical Sciences Lund, Division of Surgery, Lund, Sweden Lund University and Skåne University Hospital, Department of Clinical Sciences Lund, Division of Oncology and Pathology, Lund, Sweden University of Bristol, School of Clinical Sciences, Children's Renal Unit and Academic Renal Unit, Bristol, UK

art ic l e i nf o

a b s t r a c t

Article history: Received 9 March 2016 Received in revised form 4 July 2016 Accepted 17 July 2016

Pancreatic cancer is associated with a highly abundant stroma and low-grade inflammation. In the local tumour microenvironment, elevated glucose levels, the presence of tumour-associated stellate cells and macrophages are hypothesised to promote the tumour progression and invasion. The present study investigated the influence by the microenvironment on pancreatic cancer cell invasion in vitro. After coculture with tumour-associated pancreatic stellate cells (TPSCs), pancreatic cancer cells displayed up to 8-fold reduction in levels of epithelial-mesenchymal transition (EMT) markers E-cadherin and ZO-1, while β-catenin and vimentin levels were increased. A 3D organotypic model showed that TPSCs stimulated pancreatic cancer cell invasion, both as single cell (PANC-1) and cohort (MIAPaCa-2) invasion. The combined presence of TPSCs and M2-like macrophages induced invasion of the non-invasive BxPC-3 cells. High glucose conditions further enhanced changes in EMT markers as well as the cancer cell invasion. In summary, co-culture with TPSCs induced molecular changes associated with EMT in pancreatic cancer cells, regardless of differentiation status, and the organotypic model demonstrated the influence of microenvironmental factors, such as glucose, stellate cells and macrophages, on pancreatic cancer cell invasion. & 2016 Elsevier Inc. All rights reserved.

Keywords: Pancreatic cancer Pancreatic stellate cells Macrophages Epithelial-mesenchymal transition Invasion 3D organotypic model

1. Introduction Pancreatic cancer is characterised by a dense desmoplastic stroma, which may constitute the majority of the tumour mass. The stroma creates a tumour permissive niche and supports tumour progression, local invasion, metastasis and act as a treatment barrier [1–3]. It is now widely accepted that the pancreatic stellate cells, myofibroblast-like cells in the pancreatic microenvironment, are responsible for this stromal reaction due to their ability to

Abbreviations: ECM, extra-cellular matrix; EMT, epithelial-mesenchymal transition; TAM, tumour-associated macrophage; TGF-β, transforming growth factor β; TPSC, tumour-associated pancreatic stellate cell; ZEB1, zinc finger E-box binding homeobox 1; ZO-1, zona occludens protein 1/tight junction protein 1 n Correspondence to: Div. Oncology & Pathology, Kampradhuset, Barngatan 2B, 22185 Lund, Sweden. E-mail addresses: [email protected] (E. Karnevi), [email protected] (A.H. Rosendahl), [email protected] (K.S. Hilmersson), [email protected] (M.A. Saleem), [email protected] (R. Andersson).

produce large quantities of ECM and influence the composition and turnover of ECM proteins [4–6]. In addition, they may promote the pathogenesis of pancreatic cancer by creating a favourable tumour microenvironment through secretion of growth- and angiogenic factors as well as stimulating cancer cell migration [7–9]. Although pancreatic stellate cells have an established role in the desmoplastic stroma, further investigations are needed to elucidate the influence by paracrine communication and cell-cell interactions with the tumour microenvironment on invasive properties of the cancer cells. In order for cancer cells to disseminate and metastasise, the epithelial phenotype and cell-cell adhesions are down-regulated in favour of a more mesenchymal-like phenotype. This process of epithelial-mesenchymal transition (EMT) has been acknowledged as an intrinsic part of cancer progression [10–12]. The most significant events are down-regulation of E-cadherin, upregulation of transcription factors slug and snail, and induced expression of vimentin and N-cadherin. With the complex microenvironment around this highly invasive tumour type, further understanding

http://dx.doi.org/10.1016/j.yexcr.2016.07.017 0014-4827/& 2016 Elsevier Inc. All rights reserved.

Please cite this article as: E. Karnevi, et al., Impact by pancreatic stellate cells on epithelial-mesenchymal transition and pancreatic cancer cell invasion: Adding a third dimension in vitro, Exp Cell Res (2016), http://dx.doi.org/10.1016/j.yexcr.2016.07.017i

E. Karnevi et al. / Experimental Cell Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎

2

regarding the local interplay between present cell types and soluble factors is needed to elucidate key mechanisms and interactions for potential future targets. Pancreatic cancer is associated with low grade inflammation and one of the major risk factors for pancreatic cancer is chronic pancreatitis [13,14]. The local inflammation involves a diverse leucocyte population of which a subtype of macrophages, commonly classified as tumour-associated macrophages (TAMs) or M2-like macrophages, is a major component. Macrophages are generally divided into two subtypes: classically activated inflammatory M1 macrophages and alternatively activated M2 macrophages [15,16]. However, due to their plasticity, it is now recognised that macrophages adjust to each unique local microenvironment and the M1/M2 paradigm are merely ends of a phenotype spectrum [17,18]. TAMs are frequently found at the tumour periphery where they supply matrix remodelling enzymes and stimulate local invasion. TAMs may also play a role in EMT to facilitate dissemination of tumour cells, and in regulating the local inflammation to block anti-tumour responses [19–21]. In addition to inflammation, pancreatic cancer is commonly linked to type 2 diabetes [22,23]. Diabetes and obesity have both been recognised among the top modifiable risk factors for pancreatic cancer [24,25]. At the time of diagnosis, up to 80% of cancer patients demonstrate reduced glucose tolerance or new onset type 2 diabetes, with elevated blood glucose levels as a consequence [23,26]. Hyperglycaemia may both fuel tumour cell proliferation as well as impair response to treatment [27]. However, its role in tumour cell invasiveness remains poorly understood. The complex interplay within the tumour microenvironment has been difficult to study through conventional 2D models in vitro. In the present study, a 3D organotypic model was established to mimic the complex tumour microenvironment and cellular interactions present with the hypothesis that non-malignant cell types in the stroma promote an invasive phenotype of pancreatic cancer cells, enabling them to invade surrounding tissue. In order to test this hypothesis, the study aims to investigate the individual, as well as joint, influence by pancreatic stellate cells, M2-like macrophages and hyperglycaemia on pancreatic cancer cell invasion in vitro. In addition, co-culture studies were performed to study the influence by paracrine signalling on EMT in pancreatic cancer cells.

2. Methods 2.1. Materials All materials and reagents were purchased from Sigma-Aldrich unless stated otherwise (St Louis, MO, USA). Cell culture medium, foetal bovine serum, penicillin-streptomycin, insulin-transferrinselenium (ITS) and L-glutamine were obtained from Invitrogen (Paisley, UK). Antibody specifications, concentrations and suppliers are listed in Table 1.

2.2. Cell culture The moderately differentiated BxPC-3, and poorly differentiated PANC-1 and MIAPaCa-2 human pancreatic cancer cell lines were purchased from ATCC-LGC Standards (Manassas, VA, USA). The cells were maintained in RPMI1640 or DMEM medium supplemented with 10% foetal bovine serum (FBS) and antibiotics (100 U/ml penicillin and 100 mg/ml streptomycin) in a humified 5% CO2 atmosphere at 37 °C. The monocytic cell line THP-1 was purchased from ATCC and maintained in RPMI1640 medium supplemented with 10% FBS, 0.1% 2-mercaptoethanol and antibiotics as above. The tumour-derived pancreatic stellate cells (TPSCs) are conditionally immortalised human stellate cells generated from a pancreatic ductal adenocarcinoma specimen as previously described [28]. Briefly, cells were isolated by the outgrowth method and transformed with a temperature-sensitive SV40 large T-antigen and human telomerase. The TPSCs were evaluated for stellate cell-associated characteristics prior and after immortalisation to verify the cell line to consist of solely pancreatic stellate cells [28]. The TPSCs were maintained at 33 °C where proliferation was sustained. Prior to experiments, cells were grown to 80% confluence and transferred to 37 °C for seven days to inactivate the SV40LT and resume their primary phenotype. The cells were maintained in RPMI1640 medium supplemented with 10% FBS, 1% ITS and antibiotics as above. All experiments were performed in glucose-free RPMI1640 or DMEM supplemented with 5 mM (normal) or 25 mM (high) D-glucose, 2 mM L-glutamine, and antibiotics as above (normal or high glucose medium; NGM or HGM), unless stated otherwise.

Table 1 Antibody specifications and suppliers. Primary antibody

Species

Dilution WB

Dilution IHC/IF

Supplier

Clone/catalogue no.

GAPDH β-actin Vimentin E-cadherin β-catenin ZO-1 Claudin-1 Slug ZEB1 ZO-1 Cytokeratin Ki-67

Mouse Rabbit Rabbit Rabbit Rabbit Rabbit Rabbit Rabbit Rabbit Rabbit Mouse Mouse

1:2000 1:1000 1:1000 1:1000 1:1000 1:1000 1:1000 1:1000 1:1000 N/A N/A N/A

N/A N/A 1:400 1:200 1:100 N/A N/A N/A N/A 1:200 1:100 1:100

Millipore Cell signalling Cell signalling Cell signalling Cell signalling Cell signalling Cell signalling Cell signalling Cell signalling Sigma-Aldrich Dako Dako

Clone 6C5 #4967 #5741 #3195 #8480 #8193 #4933 #9585 #3396 #HPA001636 #M3515 Clone MIB-1

Secondary antibody

Species

Dilution WB

Dilution IHC/IF

Supplier

Catalogue no

HRP-conjugated anti-mouse HRP-conjugated anti-rabbit Alexa fluor 555 anti-rabbit Biotinylated anti-mouse Biotinylated anti-rabbit

Goat Goat Donkey Horse Goat

1:10000 1:2000 N/A N/A N/A

N/A N/A 1:200 1:200 1:200

Sigma-Aldrich Sigma-Aldrich Invitrogen Vector Vector

#A4416 #A0545 #A31572 #BA-2000 #BA-1000

Please cite this article as: E. Karnevi, et al., Impact by pancreatic stellate cells on epithelial-mesenchymal transition and pancreatic cancer cell invasion: Adding a third dimension in vitro, Exp Cell Res (2016), http://dx.doi.org/10.1016/j.yexcr.2016.07.017i

E. Karnevi et al. / Experimental Cell Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎

2.3. M2-like macrophage differentiation The THP-1 cells were seeded in T-25 flasks (7.5
105) in NGM or HGM supplemented with 10% FBS and left overnight. The cells were then incubated with 60 ng/ml phorbol 12-myristate 13acetate (PMA) for 24 h, followed by 48 h incubation with 60 ng/ml PMA, 20 ng/ml IL-4 and 20 ng/ml IL-13. M2-like macrophage differentiation was verified morphologically under microscope and by multicolour flow cytometry (CD11c (clone B-ly6), CD14 (clone M5E2), CD206 (clone 19.2) and HLA-DR (clone G46–6); BD Biosciences, San Diego, CA, USA) using a FACSCalibur (BD Biosciences) dual laser flow cytometer with Cell Quest Pro software (BD Biosciences) and analysed using FlowJo software. 2.4. Co-culture The PANC-1, MIAPaCa-2 and BxPC-3 cell lines were seeded in 6-wells plates (4
105 cells/well) and co-cultured for 48 h with or without TPSCs or M2-like macrophages seeded in inserts (3 mm pores; 8
105 cells/insert). Alternatively, pancreatic cancer cells were incubated with or without 5 ng/ml TGF-β for 48 h. The cancer cells were then harvested and transferred to inserts (8 mm pores) with NGM or HGM supplemented with 10% FBS in the lower chamber. After 6 h incubation, cancer cell lysates were collected and stored at  80 °C. 2.5. Western immunoblotting Samples were prepared and Western immunoblotting performed as outlined previously [27]. Membranes were incubated overnight at 4 °C with anti-human slug, claudin-1, ZO-1, ZEB-1, E-cadherin, vimentin or β-catenin primary antibodies. β-actin (BxPC-3) or GAPDH (PANC-1 and MIAPaCa-2) were used as loading controls. The following day membranes were washed, incubated with HRP-conjugated secondary antibody for 1 h and visualised using SuperSignal West Femto Maximum Sensitivity Substrate (ThermoFisher Scientific, Waltham, MA, USA). Immunoblotted proteins were detected using LI-COR Odyssey system with ImageStudio software (LI-COR Biosciences, Ltd., Cambridge, UK) and quantified using ImageJ software (NIH, USA). 2.6. 3D organotypic assay The three-dimensional (3D) organotypic model was set up as described by Moutasim et al. and Froeling et al. [29,30]. To make the gel, a mixture of 3.5 parts Collagen I (Millipore, Temecula, CA, USA), 3.5 parts Matrigel (Corning, Tewksbury, MA, USA), 1 part DMEM 10X, 1 part FBS and 1 part cell suspension was added to 24well plates (1 ml/well) and incubated for 1 h at 37 °C. The cell suspension contained 2.5
105 TPSCs with or without 2.5
105 M2-like differentiated macrophages in NGM or HGM supplemented with 10% FBS. After the gels had polymerised, medium was added on top and the gels left overnight at 37 °C. Cancer cells (5
105) mixed with TPSCs (2.5
105) were then seeded on top of the gels and incubated 24 h. On day three, the gels were raised by inserting metal grids in 6-wells plates with collagen-coated nylon sheets placed on top. The gels were transferred to the nylon-sheets and medium added to the well up to the bottom of the gel to create an air-liquid interface. The medium was changed every 2–3 days. After 7 or 14 days incubation, the gels were bisected, formalin fixed and paraffin embedded. 2.7. Histological and immunohistochemical analyses 2.7.1. Staining procedures Paraffin-embedded organotypic gels were sectioned 5 mm

3

thick, deparaffinised and rehydrated. For histological evaluation, sections were stained with hematoxylin and eosin (H&E) or Masson's trichrome according to the manufacturer's instructions. Prior to immunohistochemical stainings, sections were subjected to antigen retrieval, either using citrate buffer (pH 6) or Tris-EDTA buffer (pH 9), for 10 min at 90 °C in waterbath. The organotypic sections were subsequently blocked with 5% horse serum, avidin/biotin and incubated with anti-human Ki-67 or cytokeratin primary antibodies overnight at 4 °C. The following day, sections were incubated with HRP-conjugated secondary antibody for 1 h, and visualised using Vectastain Elite ABC kit and DAB (Vector, Peterborough, UK). Prior to immunofluorescence staining, antigen retrieval was performed using citrate buffer as above. The sections were subsequently blocked with 5% donkey serum, and incubated with anti-human β-catenin, E-cadherin, vimentin or ZO-1 primary antibodies overnight at 4 °C. The next day, sections were incubated with Alexa Fluors 555-conjugated secondary antibody for 1 h. The sections were washed and nuclei counterstained with DAPI. The organotypic sections were photographed at 20
magnification with CellSens Dimension software. 2.7.2. Quantification Whole slide digital images were captured at 40
magnification using Aperio ScanScope slide scanner (Aperio Technologies, Leica Biosystems, Nussloch, Germany). Total gel width, thickness of tumour epithelium and distance covered by invading cells were analysed with ImageScope software (Leica Biosystems). Six measurements of each gel section were performed to obtain a representative mean and account for variation across the gel. 2.8. Statistical analyses Western immunoblot data were analysed by two-way ANOVA with Bonferroni post-hoc test using GraphPad prism software. All graphs are expressed as mean 7 SE of a minimum of three individual experiments. Statistical analyses of organotypic sections were performed using unpaired two-tailed Student's t-test using Excel 2010 software. Graphs represent mean 7SE of six measurements across the whole gel section of depth of invasion in relation to total gel width. A P-value of o0.05 was considered statistically significant.

3. Results 3.1. TPSCs induced changes associated with EMT in pancreatic cancer cells To investigate the impact of paracrine signalling by stellate cells on EMT, the BxPC-3, PANC-1 and MIAPaCa-2 pancreatic cancer cells were cultured in the presence of TPSCs without direct cell contact. The moderately differentiated BxPC-3 cells displayed basal expression of E-cadherin and the tight junction protein ZO-1, both of which decreased by 15% (P o0.05) and 4-fold (P o0.001), respectively, in response to TPSCs (Fig. 1A). In addition, TPSCs stimulated a 2-fold increased expression of the transcription factor slug (P o0.05). Similarly, the poorly differentiated PANC-1 cells displayed 8-fold (P o0.001) and 2-fold (P o0.05) reduced levels of E-cadherin and ZO-1, respectively, after TPSC co-culture (Fig. 1A), of which ZO-1 was further reduced in the presence of high glucose levels (P o0.05). PANC-1 also showed almost 2-fold increased levels of the mesenchymal marker vimentin following co-culture (P o0.001). Lastly, MIAPaCa-2, the poorly differentiated and most mesenchymal and spindle-like cells in culture, lacked basal E-cadherin expression but demonstrated strong expression of

Please cite this article as: E. Karnevi, et al., Impact by pancreatic stellate cells on epithelial-mesenchymal transition and pancreatic cancer cell invasion: Adding a third dimension in vitro, Exp Cell Res (2016), http://dx.doi.org/10.1016/j.yexcr.2016.07.017i

4

E. Karnevi et al. / Experimental Cell Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Fig. 1. Modulation of EMT markers in three pancreatic cancer cell lines. Representative images of Western immunoblotting of claudin-1, slug, vimentin, β-catenin, E-cadherin, ZO-1 and ZEB1 in cell lysates of BxPC-3, PANC-1 and MIAPaCa-2 after 48 h co-culture with TPSCs (A) or 48 h exposure to 5 ng/ml TGF-β (B) under normal (5 mM) or high (25 mM) glucose levels. β-actin (BxPC-3) or GAPDH (PANC-1, MIAPaCa-2) were used as loading controls. Graphs represent quantification of a minimum of three individual experiments with results presented as mean7 SE fold change of expression after co-culture or TGF-β exposure, relative to control (single culture). Statistical analyses testing the significant difference between co-culture and control for both glucose levels were performed using two-way ANOVA with Bonferroni post-hoc test. N. D. ¼not detected.

Please cite this article as: E. Karnevi, et al., Impact by pancreatic stellate cells on epithelial-mesenchymal transition and pancreatic cancer cell invasion: Adding a third dimension in vitro, Exp Cell Res (2016), http://dx.doi.org/10.1016/j.yexcr.2016.07.017i

E. Karnevi et al. / Experimental Cell Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎

5

Fig. 2. The invasive properties of pancreatic cancer cells using a 3D organotypic model, Histological analyses of gel sections from organotypic models of BxPC-3, PANC-1 or MIAPaCa-2 after 7 or 14 days culture in normal (5 mM) or high (25 mM) glucose conditions. (A) H&E staining of organotypic models with or without TPSCs, (B) Trichrome staining displaying collagen fibre in blue, (C) graphs represent quantification of depth of invasion (% of total gel width) at day 7 or 14 of gels with TPSCs (empty bars) or without TPSCs (filled bars) as mean 7 SE of six measurements across each gel section. Representative images (20
magnification) from one of three independent experiments are shown. Scale bar ¼50 mm.

Please cite this article as: E. Karnevi, et al., Impact by pancreatic stellate cells on epithelial-mesenchymal transition and pancreatic cancer cell invasion: Adding a third dimension in vitro, Exp Cell Res (2016), http://dx.doi.org/10.1016/j.yexcr.2016.07.017i

6

E. Karnevi et al. / Experimental Cell Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎

vimentin, which was further increased by TPSC-derived factors (P o0.01; Fig. 1A). In addition, elevated ZO-1 (P o0.01) and β-catenin (Po 0.05) levels were observed following TPSC co-culture, and more pronounced under high glucose conditions. The TPSC-induced EMT changes were similar to or greater than those by the known EMT-inducer TGF-β (Fig. 1A and B). These results demonstrated that soluble factors and paracrine signalling by TPSCs induced molecular changes associated with EMT in pancreatic cancer cells of diverse differentiation status. 3.2. TPSCs enhanced invasion by PANC-1 and MIAPaCa-2 cells In order to mimic the tumour microenvironment and investigate the influence by cell-cell communication between pancreatic cancer cells and TPSCs on cancer cell invasion, a 3D organotypic model was established. No signs of invasion was observed for BxPC-3 cells after 7 or 14 days culture, regardless of the presence of pancreatic stellate cells or glucose levels (Fig. 2A and C). In contrast, PANC-1 and MIAPaCa-2 cells displayed induced invasion at day 7 in the presence of TPSCs, which was 28–31% (P o0.01) and 11–14% (P o0.05) higher than compared with control gels without TPSCs, respectively (Fig. 2A, C). The invasion was further increased with 20–40% by day 14 (P o0.01), and under higher glucose levels. On average, the cell lines' invasive front was seen at the depth of 58% of the total gel width, compared with 43% at day 7. The two cell lines displayed different manners of invasion, where PANC-1 cells invaded the gel as single cells or small cell clusters (o5 cells) and MIAPaCa-2 displayed cohort invasion. The invasion was further verified by Trichrome staining of collagen fibres between invading cells and the tumour cell epithelium (Fig. 2B), and cytokeratin staining of epithelial cells (Fig. S1). 3.3. Invading cell types displayed high levels of vimentin and lacked E-cadherin Associations between invasion pattern and EMT was further studied and the organotypic gel sections in the presence of TPSCs were evaluated for cellular localisation of the EMT markers E-cadherin, β-catenin, vimentin and ZO-1. No changes in EMT marker levels were observed between days 7 and 14. The noninvasive and vimentin negative BxPC-3 cells displayed membranous E-cadherin and β-catenin expression throughout the tumour epithelium, with a weak membranous ZO-1 expression at the leading edge (data not shown). PANC-1 cells in contrast, demonstrated distinct vimentin expression in both the tumour epithelium and in invading cells (Fig. 3). The vimentin intensity was more pronounced in high glucose conditions. A diffuse cytosolic E-cadherin and ZO-1 expression, and sparse membranous β-catenin expression was observed in a limited number of cells (Fig. 3). Similarly, the vimentin-positive MIAPaCa-2 tumour epithelium lacked E-cadherin expression but displayed sparse β-catenin and diffuse cytosolic ZO-1, with stronger intensity at the invasive front (data not shown). 3.4. M2-like macrophages facilitated invasion of the epithelial-like BxPC-3 pancreatic cancer cells The joint effects of TPSCs and M2-like macrophages in the microenvironment on tumour cell invasion were investigated next. After 14 days organotypic cultures in the presence of both M2-like macrophages and TPSCs, invasion of the previously non-invasive BxPC-3 cells was induced, compared with TPSC alone (Fig. 4A; P o0.05). The BxPC-3 tumour epithelium made up for less than 20% of the total gel width, both with and without the presence of M2-like macrophages, while the induced invasion was displayed at the average depth of 40% of the total gel width. The BxPC-3 cells

Fig. 3. Distribution of EMT markers in tumour cell epithelium and invading cells in 3D organotypic model, Immunofluorescent staining of EMT markers vimentin, E-cadherin, β-catenin and ZO-1 in gel sections from organotypic model of moderately invasive PANC-1 after 14 days culture in normal (5 mM) or high (25 mM) glucose. Nuclei were counterstained with DAPI. Representative images (20
magnification) from one of three independent experiments are shown. Scale bar¼ 50 mm.

displayed characteristics of cohort invasion and the invasion was further enhanced by up to 40% in high glucose conditions (P o0.05). No significant difference in invasion was observed for PANC-1 and MIAPaCa-2 cells in the presence of M2-like macrophages in this model, compared with TPSCs alone (Fig. 4B). 3.5. The presence of M2-like macrophages was associated with expression of mesenchymal markers in invading BxPC-3 cells To study if cell-cell interactions and the combined presence of macrophages and pancreatic stellate cells influenced EMT in invading cells, the cellular expression of selected EMT markers were evaluated. The presence of M2-like macrophages stimulated invasion with concomitant changes in EMT in invading BxPC-3 cell cohorts after 14 days, but not in the tumour epithelium, compared with TPSCs alone (Fig. 5). The BxPC-3 cells displayed loss of both E-cadherin and β-catenin in invading cell populations compared with the tumour cell epithelium (Fig. 5). In addition, a moderate expression of vimentin was induced in invading groups of BxPC-3 cells in the presence of M2-like macrophages. ZO-1 expression was observed at the leading edge in the tumour epithelium and was considerably enriched in invading cell cohorts (Fig. 5). The expression of the EMT markers were similar between low 5 mM and high 25 mM glucose levels. No differences in EMT markers were observed in PANC-1 and MIAPaCa-2 cells in the presence of M2like macrophages compared with TPSCs alone (data not shown).

Please cite this article as: E. Karnevi, et al., Impact by pancreatic stellate cells on epithelial-mesenchymal transition and pancreatic cancer cell invasion: Adding a third dimension in vitro, Exp Cell Res (2016), http://dx.doi.org/10.1016/j.yexcr.2016.07.017i

E. Karnevi et al. / Experimental Cell Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎

7

Fig. 4. Influence by M2-like macrophages on pancreatic cancer cell invasion, H&E staining of gel sections from organotypic models of (A) BxPC-3, (B) PANC-1 or MIAPaCa-2 cells after 14 days culture in the presence of TPSCs, with or without M2-like macrophages, under normal (5 mM) or high (25 mM) glucose conditions. Graph display quantification of depth of invasion (% of total gel width) at day 7 and 14 of BxPC-3 gels with M2-like macrophages (empty bars) or TPSCs alone (filled bars) as mean7 SE of six measurements across each gel section. Representative images (20
magnification) from one of three individual experiments are shown. Scale bar ¼50 mm.

3.6. Macrophage-derived factors stimulate changes in EMT-associated markers Finally, the contribution of paracrine signalling and soluble mediators by M2-like macrophages on tumour cell EMT was investigated through indirect co-culture between the two cell types. M2-like macrophages did not induce changes in the investigated EMT-markers to the same extent as TPSCs (Fig. 6A). Macrophagederived factors reduced the levels of slug in BxPC-3 (P o0.05), while no changes were seen for E-cadherin or ZO-1. PANC-1 however, displayed decreased E-cadherin levels by 25%, but also 2-fold reduced vimentin levels (both Pso0.01). The reduction in vimentin was further enhanced by high glucose conditions (P o0.05). ZO-1 was increased in both PANC-1 and MIAPaCa-2 cells.

4. Discussion The major non-malignant cellular components of the tumour microenvironment, the pancreatic stellate cells and tumour-associated macrophages, have been suggested to be involved in the progression of pancreatic cancer by creating a favourable tumour microenvironment and stimulate invasion, tumour growth, and development of metastases (8, 9, 19, 20). The present study utilised the benefits of a 3D organotypic model to study the joint influence by pancreatic stellate cells and M2-like macrophages on the invasive characteristics of pancreatic cancer cells. Our results showed TPSC-derived factors to stimulate changes associated with EMT in pancreatic cancer cells of diverse differentiation status, and the combined presence of M2-like macrophages and TPSCs to

induce invasion by the more epithelial-like BxPC-3 cells. In this study, co-culture with tumour-associated pancreatic stellate cells (TPSCs) stimulated changes in EMT-associated proteins for all cell lines investigated. The cancer cells showed decreased E-cadherin levels, upregulation of vimentin as well as reduced levels of the tight junction protein ZO-1. These results are in line with previous studies where pancreatic stellate cells or cancer-associated fibroblasts have been shown to potentially induce EMT in pancreatic cancer cells, and thereby facilitate invasion [31,32]. Furthermore, the present study demonstrates TPSCs to induce EMT-associated changes by paracrine communication, regardless of differentiation status. It has been hypothesised that TPSCs potentially exert some of their tumour promoting effects through TGF-β, a known inducer of EMT. Here, the changes displayed after co-culture with TPSCs were similar to the effect of TGF-β treatment. This indicates that TGF-β may at least in part contribute to the TPSC-induced changes in EMT observed. However other factors secreted by TPSCs such as uPA, FGF and PDGF may likely also be involved [28]. It has been shown that cancer-associated fibroblasts or TPSCs can stimulate migration or invasion of cancer cells in vitro, although mainly through 2D models [8,31]. The organotypic model has previously displayed favourable results for visualisation of cancer cell invasion and interplay with the local microenvironment [32,33]. The results here demonstrate cell-cell interactions with TPSCs to stimulate different manners of invasion: single cell invasion (PANC-1) and cohort invasion (MIAPaCa-2). The pancreatic cancer cells also displayed significantly increased invasion in the presence of TPSCs, compared with gels lacking stellate cells. The epithelial-like BxPC-3 cells, on the other hand, remained noninvasive with an intact tumour epithelium. The cell lines’ invasive

Please cite this article as: E. Karnevi, et al., Impact by pancreatic stellate cells on epithelial-mesenchymal transition and pancreatic cancer cell invasion: Adding a third dimension in vitro, Exp Cell Res (2016), http://dx.doi.org/10.1016/j.yexcr.2016.07.017i

8

E. Karnevi et al. / Experimental Cell Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Fig. 5. Distribution of EMT markers in tumour cell epithelium and invasive cells in the presence of both TPSCs and M2-like macrophages, Immunofluorescent staining of EMT markers vimentin, E-cadherin, β-catenin and ZO-1 in gel sections from organotypic models of BxPC-3 pancreatic cancer cells after 14 days in the presence of TPSCs, with or without M2-like macrophages, under normal (5 mM) or high (25 mM) glucose conditions. Nuclei were counterstained with DAPI. Representative (20
magnification) from one of three independent experiments is shown. Scale bar ¼50 mm.

behaviour displayed in the organotypic model corresponds to changes in EMT markers visualised by Western blot. This strengthens the pro-tumour role of TPSCs in the tumour microenvironment, and the evidence of paracrine communication and cell-cell interactions with pancreatic cancer cells. Furthermore, it suggests TPSCs to be an important target with the potential to regulate invasion and the associated EMT. Previous studies have focused on other pro-tumour properties of stellate cells in order to inhibit the tumourigenesis. The desmoplastic stroma produced by pancreatic stellate cells has been shown to create a treatment barrier around the tumour, limiting bioavailability of drugs [1–3]. In addition, ECM components in the microenvironment have been shown to influence cancer cell migration and dissemination [34]. To surpass this barrier, Olive et al. increased drug availability and chemotherapy response by targeting the pancreatic stellate cells and thereby reducing the stromal compartment before administrating treatment [35]. There has however been other studies demonstrating anti-tumour influences by the stroma, whereby deleting the stromal compartment results in increased aggressiveness of pancreatic tumours and reduced survival [36,37]. Based on the conflicting results by targeting the stroma, future investigative studies to reduce tumour aggressiveness and/or inhibit tumour progression stimulated by pancreatic stellate cells will require careful design in order to solely target the pro-tumour properties.

In the local pancreatic tumour microenvironment, tumour-associated macrophages (TAMs) play a central role next to the pancreatic stellate cells. In a study with squamous carcinoma cells, M2 macrophages together with fibroblasts enhance the invasion rate of cancer cells compared to fibroblasts alone [33]. In a similar manner, the combined presence of TPSCs and M2-like macrophages in the present study induced invasion of more epitheliallike BxPC-3 pancreatic cancer cells, which displayed no invasion in the presence of TPSCs alone. Interestingly, the BxPC-3 cells disrupted the previously intact tumour epithelium and displayed invasion with an associated loss of E-cadherin expression. Staining of EMT markers displayed expression patterns resembling the more poorly differentiated PANC-1 and MIAPaCa-2 cells, in particular in the invaded groups of BxPC-3 cells. When investigating the influence on EMT-associated markers after co-culture, the M2like macrophages induced changes that differed from those demonstrated by TPSCs-derived factors, such as reduced vimentin levels and increased ZO-1. This may be due to differences in secreted factors and thereby the paracrine signalling pathways involved. The M2-like macrophages most likely stimulate invasion and changes in EMT-markers by secretion of cytokines, MMPs and growth factors, which all are associated with invasion, metastasis and ECM degradation [38–42]. In studies investigating colon cancer and breast cancer, a chemotactic loop between cancer cells and macrophages have been observed where the macrophages

Please cite this article as: E. Karnevi, et al., Impact by pancreatic stellate cells on epithelial-mesenchymal transition and pancreatic cancer cell invasion: Adding a third dimension in vitro, Exp Cell Res (2016), http://dx.doi.org/10.1016/j.yexcr.2016.07.017i

E. Karnevi et al. / Experimental Cell Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎

9

Fig. 6. Influence by macrophage-derived factors on EMT-associated markers, Representative images of Western immunoblotting of claudin-1, slug, vimentin, β-catenin, E-cadherin, ZO-1 and ZEB1 in cell lysates of BxPC-3, PANC-1 and MIAPaCa-2 after 48 h co-culture with M2-like macrophages. β-actin (BxPC-3) or GAPDH (PANC-1 and MIAPaCa-2) were used as loading controls. Graphs represent quantification of a minimum of three independent experiments with data presented as mean 7 SE fold change of expression after co-culture, relative to control (single culture). Statistical analyses testing the significant difference between co-culture and control for both glucose levels were performed using two-way ANOVA with Bonferroni post-hoc test. N.D. ¼not detected.

stimulate cancer cell invasion by secretion of EGF [20,38]. The interplay between TPSCs and macrophages themselves is also of interest, since pancreatic stellate cells have been suggested to contribute to local inflammation and recruitment of inflammatory cells [43,44]. Erez et al. investigated inflammatory profiles of cancer-associated fibroblasts in a wide range of cancers, and found similar signatures comparing pancreatic cancer to the inflammatory environments present in skin cancer [44]. In conclusion, TPSCs induced changes associated with EMT in human pancreatic cancer cells and stimulated both single cell invasion (PANC-1) and cohort invasion (MIAPaCa-2). This suggests the inter-cell communication with pancreatic stellate cells to be central for pancreatic cancer cell invasion, and an important potential target to restrict tumour spreading. The presence of M2like macrophages stimulated the invasion further as well as induced invasion of more epithelial-like and non-invasive cells. This demonstrates the importance of the tumour microenvironment in stimulating cancer progression and cancer cell invasion into surrounding tissue and opens up for further investigations in the search of new targets outside the tumour itself in order to inhibit the tumour progression.

5. Funding This study was supported by the South Swedish Health Care Region (Region Skåne ALF), the Skåne University Hospital Fund, the Royal Physiographic Society in Lund, The John and Augusta Persson Foundation, the Faculty of Medicine at Lund University

and the Gyllenstiernska Krapperups' Foundation.

6. Author contributions EK designed the study. EK, KS, AHR and MAS made significant contributions to the experimental design, acquisition and interpretation of data, manuscript preparation and editing. RA provided funding for the study and manuscript editing. All authors read and approved the final manuscript.

7. Competing interests No competing interests declared.

Acknowledgements We thank Professor Gareth Thomas' lab at University of Southampton, Southampton, United Kingdom for teaching us the 3D organotypic model. We thank Juan Vaz, Johannes Byrling, Adna Grahovic and Yasmine Adbul Rahim for their assistance with the experimental work. Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.yexcr.2016.07.017.

Please cite this article as: E. Karnevi, et al., Impact by pancreatic stellate cells on epithelial-mesenchymal transition and pancreatic cancer cell invasion: Adding a third dimension in vitro, Exp Cell Res (2016), http://dx.doi.org/10.1016/j.yexcr.2016.07.017i

10

E. Karnevi et al. / Experimental Cell Research ∎ (∎∎∎∎) ∎∎∎–∎∎∎

References [1] C.J. Whatcott, C.H. Diep, P. Jiang, A. Watanabe, J. LoBello, C. Sima, et al., Desmoplasia in primary tumors and metastatic lesions of pancreatic cancer, Clin. Cancer Res.: Off. J. Am. Assoc. Cancer Res. (2015). [2] M. Erkan, S. Hausmann, C.W. Michalski, A.A. Fingerle, M. Dobritz, J. Kleeff, et al., The role of stroma in pancreatic cancer: Diagnostic and therapeutic implications, Nat. Rev. Gastroenterol. Hepatol. 9 (8) (2012) 454–467. [3] R. Jaster, J. Emmrich, Crucial role of fibrogenesis in pancreatic diseases, Best Pract. Res. Clin. Gastroenterol. 22 (1) (2008) 17–29. [4] T.W. Yen, N.P. Aardal, M.P. Bronner, D.R. Thorning, C.E. Savard, S.P. Lee, et al., Myofibroblasts are responsible for the desmoplastic reaction surrounding human pancreatic carcinomas, Surgery 131 (2) (2002) 129–134. [5] M. Erkan, C. Reiser-Erkan, C.W. Michalski, S. Deucker, D. Sauliunaite, S. Streit, et al., Cancer-stellate cell interactions perpetuate the hypoxia-fibrosis cycle in pancreatic ductal adenocarcinoma, Neoplasia 11 (5) (2009) 497–508. [6] M.V. Apte, S. Park, P.A. Phillips, N. Santucci, D. Goldstein, R.K. Kumar, et al., Desmoplastic reaction in pancreatic cancer: role of pancreatic stellate cells, Pancreas 29 (3) (2004) 179–187. [7] J. Tod, V. Jenei, G. Thomas, D. Fine, Tumor-stromal interactions in pancreatic cancer, Pancreatol.: Off. J. Int. Assoc. Pancreatol. (IAP) 13 (1) (2013) 1–7. [8] R.F. Hwang, T. Moore, T. Arumugam, V. Ramachandran, K.D. Amos, A. Rivera, et al., Cancer-associated stromal fibroblasts promote pancreatic tumor progression, Cancer Res. 68 (3) (2008) 918–926. [9] A. Masamune, K. Kikuta, T. Watanabe, K. Satoh, M. Hirota, T. Shimosegawa, Hypoxia stimulates pancreatic stellate cells to induce fibrosis and angiogenesis in pancreatic cancer, Am. J. Physiol. Gastrointest. Liver Physiol. 295 (4) (2008) G709–G717. [10] J.A. Castellanos, N.B. Merchant, N.S. Nagathihalli, Emerging targets in pancreatic cancer: epithelial-mesenchymal transition and cancer stem cells, OncoTargets Ther. 6 (2013) 1261–1267. [11] B. Hotz, M. Arndt, S. Dullat, S. Bhargava, H.J. Buhr, H.G. Hotz, Epithelial to mesenchymal transition: expression of the regulators snail, slug, and twist in pancreatic cancer, Clin. Cancer Res. 13 (16) (2007) 4769–4776. [12] S. Nakajima, R. Doi, E. Toyoda, S. Tsuji, M. Wada, M. Koizumi, et al., N-cadherin expression and epithelial-mesenchymal transition in pancreatic carcinoma, Clin. Cancer Res.: Off. J. Am. Assoc. Cancer Res. 10 (12 Pt 1) (2004) 4125–4133. [13] J.B. Greer, D.C. Whitcomb, Inflammation and pancreatic cancer: an evidencebased review, Curr. Opin. Pharmacol. 9 (4) (2009) 411–418. [14] A.B. Lowenfels, P. Maisonneuve, G. Cavallini, R.W. Ammann, P.G. Lankisch, J. R. Andersen, et al., Pancreatitis and the risk of pancreatic cancer. International Pancreatitis Study Group, N. Engl. J. Med. 328 (20) (1993) 1433–1437. [15] P.J. Murray, T.A. Wynn, Protective and pathogenic functions of macrophage subsets, Nat. Rev. Immunol. 11 (11) (2011) 723–737. [16] F.O. Martinez, A. Sica, A. Mantovani, M. Locati, Macrophage activation and polarization, Front. Biosci.: J. virtual Libr. 13 (2008) 453–461. [17] B. Ruffell, N.I. Affara, L.M. Coussens, Differential macrophage programming in the tumor microenvironment, Trends Immunol. 33 (3) (2012) 119–126. [18] B.Z. Qian, J.W. Pollard, Macrophage diversity enhances tumor progression and metastasis, Cell 141 (1) (2010) 39–51. [19] C. Neyen, A. Pluddemann, S. Mukhopadhyay, E. Maniati, M. Bossard, S. Gordon, et al., Macrophage scavenger receptor a promotes tumor progression in murine models of ovarian and pancreatic cancer, J. Immunol. 190 (7) (2013) 3798–3805. [20] C.E. Green, T. Liu, V. Montel, G. Hsiao, R.D. Lester, S. Subramaniam, et al., Chemoattractant signaling between tumor cells and macrophages regulates cancer cell migration, metastasis and neovascularization, PloS One 4 (8) (2009) e6713. [21] S.M. Wormann, K.N. Diakopoulos, M. Lesina, H. Algul, The immune network in pancreatic cancer development and progression, Oncogene 33 (23) (2014) 2956–2967. [22] S.T. Chari, C.L. Leibson, K.G. Rabe, L.J. Timmons, J. Ransom, M. de Andrade, et al., Pancreatic cancer-associated diabetes mellitus: Prevalence and temporal association with diagnosis of cancer, Gastroenterology 134 (1) (2008) 95–101. [23] R. Pannala, J.B. Leirness, W.R. Bamlet, A. Basu, G.M. Petersen, S.T. Chari, Prevalence and clinical profile of pancreatic cancer-associated diabetes mellitus, Gastroenterology 134 (4) (2008) 981–987. [24] B. Bao, Z. Wang, Y. Li, D. Kong, S. Ali, S. Banerjee, et al., The complexities of obesity and diabetes with the development and progression of pancreatic

cancer, Biochim. ET Biophys. Acta 1815 (2) (2011) 135–146. [25] R. Huxley, A. Ansary-Moghaddam, A. Berrington de Gonzalez, F. Barzi, M. Woodward, Type-II diabetes and pancreatic cancer: a meta-analysis of 36 studies, Br. J. Cancer 92 (11) (2005) 2076–2083. [26] J. Permert, I. Ihse, L. Jorfeldt, H. von Schenck, H.J. Arnqvist, J. Larsson, Pancreatic cancer is associated with impaired glucose metabolism, Eur. J. Surg. – Acta Chir. 159 (2) (1993) 101–107. [27] E. Karnevi, K. Said, R. Andersson, A.H. Rosendahl, Metformin-mediated growth inhibition involves suppression of the IGF-I receptor signalling pathway in human pancreatic cancer cells, BMC Cancer 13 (2013) 235. [28] A.H. Rosendahl, C. Gundewar, K. Said Hilmersson, L. Ni, M.A. Saleem, R. Andersson, Conditionally immortalized human pancreatic stellate cell lines demonstrate enhanced proliferation and migration in response to IGF-I, Exp. Cell Res. 330 (2) (2015) 300–310. [29] K.A. Moutasim, M.L. Nystrom, G.J. Thomas, Cell migration and invasion assays, Methods Mol. Biol. 731 (2011) 333–343. [30] F.E. Froeling, J.F. Marshall, H.M. Kocher, Pancreatic cancer organotypic cultures, J. Biotechnol. 148 (1) (2010) 16–23. [31] K. Kikuta, A. Masamune, T. Watanabe, H. Ariga, H. Itoh, S. Hamada, et al., Pancreatic stellate cells promote epithelial-mesenchymal transition in pancreatic cancer cells, Biochem. Biophys. Res. Commun. 403 (3–4) (2010) 380–384. [32] F.E. Froeling, T.A. Mirza, R.M. Feakins, A. Seedhar, G. Elia, I.R. Hart, et al., Organotypic culture model of pancreatic cancer demonstrates that stromal cells modulate E-cadherin, beta-catenin, and Ezrin expression in tumor cells, Am. J. Pathol. 175 (2) (2009) 636–648. [33] N. Linde, C.M. Gutschalk, C. Hoffmann, D. Yilmaz, M.M. Mueller, Integrating macrophages into organotypic co-cultures: a 3D in vitro model to study tumor-associated macrophages, PloS One 7 (7) (2012). [34] K.V. Nguyen-Ngoc, K.J. Cheung, A. Brenot, E.R. Shamir, R.S. Gray, W.C. Hines, et al., ECM microenvironment regulates collective migration and local dissemination in normal and malignant mammary epithelium, Proc. Natl. Acad. Sci. USA 109 (39) (2012) E2595–E2604. [35] K.P. Olive, M.A. Jacobetz, C.J. Davidson, A. Gopinathan, D. McIntyre, D. Honess, et al., Inhibition of Hedgehog signaling enhances delivery of chemotherapy in a mouse model of pancreatic cancer, Science 324 (5933) (2009) 1457–1461. [36] A.D. Rhim, P.E. Oberstein, D.H. Thomas, E.T. Mirek, C.F. Palermo, S.A. Sastra, et al., Stromal elements act to restrain, rather than support, pancreatic ductal adenocarcinoma, Cancer Cell 25 (6) (2014) 735–747. [37] B.C. Ozdemir, T. Pentcheva-Hoang, J.L. Carstens, X. Zheng, C.C. Wu, T. R. Simpson, et al., Depletion of carcinoma-associated fibroblasts and fibrosis induces immunosuppression and accelerates pancreas cancer with reduced survival, Cancer Cell 25 (6) (2014) 719–734. [38] S. Goswami, E. Sahai, J.B. Wyckoff, M. Cammer, D. Cox, F.J. Pixley, et al., Macrophages promote the invasion of breast carcinoma cells via a colonystimulating factor-1/epidermal growth factor paracrine loop, Cancer Res. 65 (12) (2005) 5278–5283. [39] B. Baran, I. Bechyne, M. Siedlar, K. Szpak, B. Mytar, J. Sroka, et al., Blood monocytes stimulate migration of human pancreatic carcinoma cells in vitro: the role of tumour necrosis factor – alpha, Eur. J. Cell Biol. 88 (12) (2009) 743–752. [40] K. Kessenbrock, V. Plaks, Z. Werb, Matrix metalloproteinases: regulators of the tumor microenvironment, Cell 141 (1) (2010) 52–67. [41] Y. Kuwada, T. Sasaki, K. Morinaka, Y. Kitadai, N. Mukaida, K. Chayama, Potential involvement of IL-8 and its receptors in the invasiveness of pancreatic cancer cells, Int. J. Oncol. 22 (4) (2003) 765–771. [42] M. Lesina, M.U. Kurkowski, K. Ludes, S. Rose-John, M. Treiber, G. Kloppel, et al., Stat3/Socs3 activation by IL-6 transsignaling promotes progression of pancreatic intraepithelial neoplasia and development of pancreatic cancer, Cancer Cell 19 (4) (2011) 456–469. [43] V. Tjomsland, L. Niklasson, P. Sandstrom, K. Borch, H. Druid, C. Bratthall, et al., The desmoplastic stroma plays an essential role in the accumulation and modulation of infiltrated immune cells in pancreatic adenocarcinoma, Clin. Dev. Immunol. 2011 (2011) 212810. [44] N. Erez, M. Truitt, P. Olson, S.T. Arron, D. Hanahan, Cancer-associated fibroblasts are activated in incipient neoplasia to orchestrate tumor-promoting inflammation in an NF-kappaB-dependent manner, Cancer Cell 17 (2) (2010) 135–147.

Please cite this article as: E. Karnevi, et al., Impact by pancreatic stellate cells on epithelial-mesenchymal transition and pancreatic cancer cell invasion: Adding a third dimension in vitro, Exp Cell Res (2016), http://dx.doi.org/10.1016/j.yexcr.2016.07.017i